To address the fact that changes in network topology can have an impact on the performance of routing, this paper proposes an Elastic Routing Algorithm based on Multi-Agent Deep Deterministic Policy Gradient (ERA-MADDPG), which is implemented within the framework of Multi-Agent Deep Deterministic Policy Gradient (MADDPG) in deep reinforcement learning. The algorithm first builds a three-layer architecture based on Software-Defined Networking (SDN). The top-down layers are the multi-agent layer, the controller layer, and the data layer. The architecture’s processing flow, including real-time data layer information collection and dynamic policy generation, enables the ERA-MADDPG algorithm to exhibit strong elasticity by quickly adjusting routing decisions in response to topology changes. The actor-critic framework combined with Convolutional Neural Networks (CNN) to implement the ERA-MADDPG routing algorithm effectively improves training efficiency, enhances learning stability, facilitates collaboration, and improves algorithm generalization and applicability. Finally, simulation experiments demonstrate that the convergence speed of the ERA-MADDPG routing algorithm outperforms that of the Multi-Agent Deep Q-Network (MADQN) algorithm and the Smart Routing based on Deep Reinforcement Learning (SR-DRL) algorithm, and the training speed in the initial phase is improved by approximately 20.9% and 39.1% compared to the MADQN algorithm and SR-DRL algorithm, respectively. The elasticity performance of ERA-MADDPG is quantified by re-convergence speed: under 5–15% topology node/link changes, its re-convergence speed is over 25% faster than that of MADQN and SR-DRL, demonstrating superior capability to maintain routing efficiency in dynamic environments.

As network environments become increasingly dynamic and users’ Quality of Service (QoS) demands grow more diverse, efficient and adaptive routing strategies are urgently needed. However, traditional routing strategies suffer from limitations such as poor adaptability to fluctuating traffic, lack of differentiated service handling, and slow convergence in complex network scenarios. To this end, we propose a routing strategy based on multi-agent deep deterministic policy gradient for differentiated QoS services (RS-MADDPG) in a software-defined networking (SDN) environment. First, network state information is collected in real time and transmitted to the control layer for processing. Then, the processed information is forwarded to the intelligent layer. In this layer, multiple agents cooperate during training to learn routing policies that adapt to dynamic network conditions. Finally, the learned policies enable agents to perform adaptive routing decisions that explicitly address differentiated QoS requirements by incorporating a custom reward structure that dynamically balances throughput, delay, and packet loss according to traffic type. Simulation results demonstrate that RS-MADDPG achieves convergence approximately 30 training cycles earlier than baseline methods, while improving average throughput by 3%, reducing latency by 7%, and lowering packet loss rate by 2%.